Econometric Analysis of China-ECOWAS agricultural products trade

ABSTRACT: Agriculture trade remains the economic fulcrum of most African countries as the continent continues to host the largest percent of arable land. This research analyzed the Economic Community of West African States (ECOWAS) and China’s agricultural products trade determinants based on 19 years (2000-2018) panel dataset of West African countries aggregate agricultural products exports ($) and macroeconomic variables; GDP, population, arable land, language investment, and trade association(WTO)) as predictors. The PPML estimation method was employed due to its prediction accuracy, the size of the data, and potential hetroskadacity issues. With a 78.5% prediction power, the model explained the variation in ECOWAS-China agricultural trade (Exports). GDPj, lnPOPj, lnPOPi, and lnARLj, LndLj, ConfInsj, and WTOij were positive and statistically significant determinants of trade as hypothesized by existing trade literature. In addition, the China’s population (lnPOPj) had a value of 0.5877, which is significant at the 5% level, indicating that a 1% increase in the Chinese population significantly increases trade in agricultural products with ECOWAS states. The coefficient of distance (Dij) is -4.4573 statistically significant at the 1% level, indicating that distance between partners impedes trade flow. There are unidentified barriers that delay the progress of trade in agricultural products between ECOWAS and China. Based on the above findings, Investments in ECOWAS arable lands demand urgent attention if significant progress in exports is expected, additionally, governments of both partners should assist Agricultural research and development to identify and rectify stifling trade barriers. Furthermore, as trade between ECOWAS and China has not yet reached its full peak, studies on export determinants of individual Agro-commodities and potentials are needed to enrich literature.


INTRODUCTION
Over the last decades, Sustained interest in China-African economic ties has resulted in hundreds of media stories and opinions, dramatic assertions, and robust misconceptions, but surprisingly evidence about the reasons of the growing agricultural exports from key economic states of West Africa (ECOWAS) is limited (MIAO et al., 2020;VILLORIA, 2009; ZHANG Ciência Rural, v.53, n.1, 2023. Abdullahi et al. et al., 2010). Although, some studies (FUKASE & MARTIN, 2016;KONINGS, 2007) purport that agricultural imports from Africa as insignificant in volume as compared with China's exports to the region, CHATHAM HOUSE (2020) data shows that except for oilseeds and crude oil, import of fish aquatic resources, gums, and rubber among other agricultural products exports from the ECOWAS region exceeds $1.2bn. Figures 1 and 2 explore the overall performance of ECOWAS agricultural imports and exports whilst figure 3 shows ECOWAS imports to China for the past 18 years. Furthermore, the agricultural trade flow (export and import) and the market share is in table 1. Reaching a high peak in 2012, Agricultural exports have exhibited seasonal growth since 2000 to date. This growth does not come as a surprise since most African States are heavily reliant on Agricultural exports earnings. However, to fully understand the myths surrounding this trade relation, we draw analogies from both recent findings and evidence from early trade theologians. The traditional gravity model explains the variations in trade based on economic size and distance (TINbERGEN, J. 1962). However, with the advancement and dynamics in trade, several possible microeconomic and macroeconomic indicators have been discovered to influence trade among individuals and groups of trade partners (AbOULEZZ, 2016;AKOWUAH et al., 2020;NASRULLAH et al., 2020;NGOMA, 2020;VU et al., 2020). In the context of China and Africa bilateral trade, factors such as language investment (YEbOAH et al., 2021), WTO membership, (LIEN et al., 2019;SHAHRIAR et al., 2020) institutional quality (DIDIER & HOARAU, 2021;GOLD & RASIAH, 2021) economic agreement and trade agreements (GUAN & IP PING SHEONG, 2020) have shown significant influence on the volume and direction of trade respectively. For agricultural trade, H. SEN ZHANG et al., (2010) found out similarities and possible potentials between China and African States based on ongoing cooperation that seeks to promote agricultural trade.
Although, the aforementioned studies employed the extended gravity model in Analyzing China and Africa trade, are still missing ingredients that demands further investigation. Therefore, the present study seeks to address multiple gaps and in doing so makes a vital contribution. First, the study extends the limited work on the drivers of China -west Africa Agricultural trade using the current trade data; Secondly, no previous research to the best of the authors' knowledge and through search in the peer-reviewed database has empirically analyzed ECOWAS agricultural exports to China within the same time frame, despite the existing level of Agricultural cooperation between the two economies. Moreover, existing literature on trade determinants is only limited to the Sub-Saharan African region other than regional trade blocks (VON ESSEN, 2017) which forms the basis of our research question;  Considering the proportion of Africa's arable land resource (76%) to the rest of the world and China's growing influence in Africa, this research is of paramount interest in enriching literature and policymakers as the determinants uncovered will propel strong policy formulation regarding China and Africa future Agricultural trade policies.  Ciência Rural, v.53, n.1, 2023. Abdullahi et al. The growing population, improved route for transportation and China's language investment in ECOWAS have significant (positive) influence on the volume of trade aside from common trade association (WTO membership) as hypothesized by other trade literature. In another vein, China's population growth and arable land size present potential opportunities for increased imports from the ECOWAS region. Moreover, the geographical distance, which signifies trade barriers as reported in original gravity model literature, has similar negative repercussions per our current findings. The other sections of the research are structured as follows; the literature review and summary, Materials and methods, data analysis, Results and discussion, and conclusion.

Literature review The gravity model
The gravitational theory of trade stems its roots from the early works of Isaac Newton's gravity concept far back in 1687. The original concept, which estimated the gravity of objects, based on their Mass and the relative distance was later fused into international trade by TINbERGEN (1962) and later extended by LINNEMANN (1962). In their theory, the economic Mass of a country was represented by GDP whereas distance denoted the Geographical distance between the economies involved. Later, bECHDOT & NIEDERCORN (1969) also investigated the empirical authenticity of the gravity model in the context of utility theory. In 1979, ANDERSON (2003, derived the first equation of the gravity model by applying the product differentiation model. Since then, several confirmatory works have been done with varying outcomes. bERGSTRAND (1985, 1989, and 1990) applied the microeconomic foundations of trade through models of monopolistic competition. DEARDORFF (1995), also proved that the model was consistent with neoclassical models derived in a defective competition framework. However, VAN WIN COOP & ANDERSON (2003) disagreed that there is no theoretical basis for the estimated equations of the gravity model. notwithstanding, after many years of its application the model's efficiency still holds much validity and continues to be applied in international trade applying different modifications.

Application of gravity model in agricultural trade
Though some studies SHAKUR (2012), WANG et al., (2014) have predicted China's potential of minimizing agricultural imports based on its growing extensive production output, intrinsic and extrinsic constraints in sustainable food production coupled with population growth and changing consumer demands have rather lead to increased imports over the years.
Applying the gravity model based on 23-year panel data  On the part of Africa's Agricultural export trend, several pertinent findings concerning what influences export from individual African States have been recorded in literature. For instance, VON ESSEN (2017) reported that a 1% increase in the GDP of an SSA country (Sub-Saharan Africa) should lead to a 0.28% increase in its export of agricultural commodities to China. Similarly, a 1 % increase in infrastructure would lead to a 0.12 % increase in agricultural commodity exports to China. VON ESSEN (2017) also found that the more arable land there is, the higher the possibility of supplying more agricultural commodities. GDP, natural resource endowment, institutional quality, and infrastructure have been identified as determinants of Chinese imports from SSA countries.
NIGHT (2015) evaluated Kenya's cattle exports to international partners over 23 years using panel data . The findings showed that Kenya's GDP, importer's per capita GDP, and Kenya's per capita GDP were all major predictors of Kenya's livestock exports to global partners. AbDULLAHI et al., 2021, investigated Nigeria's cocoa exports using panel data covering 24 years and Nigeria's 36 global trading partners. Using the PPML, the results indicated that export flows of Nigerian cocoa are favorably correlated with trade association (WTO membership), exchange rate, GDP, colonial ties, and EU, while per capita GDP, distance, landlocked status, and AU have a negative correlation with exports.
EbAIDALLA & AbDALLA (2016) identified the determinants of Sudan agricultural exports with 31 global trading partners from 1995 to 2011. GDP, population size, and infrastructure play a favorable and substantial influence in increasing exports performance while distance was found negative and significant on exports performance. Moreover, bAKARI & MOHAMED (2018) observed that GDP has a weak correlation with agricultural exports.
POTELWA, LUbINGA, & NTSHANGASE (2016) assessed the elements that influence South Africa's agricultural exports to global markets using panel data from 2001 to 2014. It was revealed that, as South Africa's and importers' GDPs rise, agricultural exports rise as well. The increase of agricultural exports to its trading partners is unaffected by distance and political stability. The population of the importer and the export capacity of the exporter had a favorable impact on the growth of South Africa's agricultural exports to its trading partners.
The above studies have highlighted the determinants of China's Agricultural exports to major trading partner countries with possible determinants. In the case of Africa's exports, a significant number of individual countries exports have also been examined in both current and previous literature. However, based on the growing tides between China and Africa, which have sprouted various cooperation forums such as SADC, FOCAC among others and the controversies surrounding China-Africa trade, this present study significant in providing possible Ciência Rural, v.53, n.1, 2023. Abdullahi et al. answers. Additionally, there appears to be limited study focusing on the ECOWAS Agricultural trade with China hence this study will also prove vital in filling such literature gap. Table 2 summarizes key literature findings based on the Agricultural imports of China and exports of Africa between 1995 to 2019.

MATERIALS AND METHODS
In this study, the regression analysis according to AbDULLAHI et al., 2021 was employed using a panel data of agricultural exports from 15 West African countries. Those are Nigeria, Ghana, benin, Cote d'Ivoire, Niger, Mali, Togo, Guinea, burkina Faso, Guinea bissau Senegal, Carbo Verde, Gambia, Guinea bissau, and Sierra Leone. The countries were selected based on the continuous agricultural trade relations between China and these West African countries. Nineteen years of panel data of exports of agricultural products to China from these 15 countries were collected, starting from 2000-2018. both dependent variable and independent variable were obtained from the reputable database as elaborated in table 3 below:

Models Specification
The model of gravity elucidates the flows of trade as a log function of income and distance between countries. It forecast that bilateral trade is significantly influenced by distance (negative) and income (positive) which can be expressed mathematically as: Export ij = Exports flow from country j to i, G j and G i = GDP per capita of both countries, whilst D ij = geographical distance between country i and j.
The linear representation of the model is as follows: LnExport ij = ɑ + β 1 logG і + β 2 logG j + β 3 logD іj (2) According to the generalized gravity model of trade, the volume of exports between two countries, (Exports ij ), is a function of their GDPs, populations, and distance, population, and other set of dummy variables that either help or hinder trade between two countries.
Where Exports ij means exports flow from country i to j, G i and G j represent GDP per capita of both countries, P i and P j denotes population of country i and j, D ij represents their geographical distance between the nearest port of the two countries, V ij represents other variables that may influence agricultural exports. ε ij means error term, β's are the model parameters.
The PPML model for this research work is expressed as: (5) Where Export ij stands for total exports of agricultural products from ECOWAS members to China from 2000 to 2018 signifying our dependent variable. The independent variables were elucidated as follows: Ln (GDP i x GDP j ) stands for the GDP value of the trading partners, which shows the size of ln (WTO ijt ) should have expected positive signs while ln (Landlocked j ) is identified as an impediment factor of trade expected to have a negative sign. The Heckman selection model is made up of two equations: sample selection (eq. 6, 7) and outcome selection (eq. 8). The sample selection model is as follows: Where t * ijt represents a latent variable and it is not observed but we do observe if countries trade or not, such that t ijt = 1 if t * ijt > 0 and t * ijt = 1, if t ijt = 0 and denotes a vector variable that affects t * ijt . µ ijt is the error term. Apart from the above-mentioned variables, other variables ijt may influence t * ijt in this study. The study has included certain dummies in addition to the other independent variables to see how the Chinese Confucius institutions, landlocked countries, and WTO membership affect agricultural products exports. Selection model: t * ijt = ƞ0 + ƞ1ln(GDP * GDP) + ƞ2ln(POPi + POPj) + ƞ3lnDij + ƞ4lnEXCij + ƞ5lnARLj + ƞ6InfraSj + ƞ7Lndlij + ƞ8Conf Insij + ƞ9WTOij + µ ijt (7) Outcome model: Lnexportsij = ɑ 0 + β 1 ln(GDP * GDP) + β 2 ln(POP i + POP j ) + β 3 lnD ij + β 4 lnEXC ij + β 5 lnARL j + β 6 InfraS j + γ 1 Lndl ij + γ 2 Conf Ins ij + γ 3 WTO ij + ε ij (8) In econometrics, independent variables selection is a challenging task. AMEMIYA, (1980) states that the selection of regression analysts should be based on economic theory as well as statistical logic. In the estimations of the econometric model, the omitted variables may lead to biased and incorrect conclusions (WOOLDRIDGE, 2002). Model misspecifications can be caused by two factors: (1) incorrect functional form, and (2) invalid assumptions on the distribution of the disturbance term (bERA & JARQUE, 1982). Moreover, we must consider the model's correct specification, functional forms, and regressors. We selected the relevant variables for the specification of the empirical gravity model based on the above principles and instructions, as well as previous empirical studies and trade theories.

Descriptive statistics and test of multicollinearity
based on the summary descriptive statistics from table 4, we obtained an overview of the variables presented in the study and examined data normality before the PPML estimation. On average ECOWAS exports $30.47 worth of Agricultural commodities to the Chinese territory between 2000 and 2019 with the highest and lowest trade volume of 37.5 and 25.4 respectively. Although, the current volume of exports is less than 1% of the global share of agricultural products trade, the average volume far exceeds the total ECOWAS exports of the year 2000 (CHATHAM HOUSE, 2021). This makes it worthwhile studying the contributing factors enhancing this trade. Similarly, the average performance of economic growth indicators such as GDP and population of both ECOWAS and China reveals a potential growth of agricultural trade as elaborated in table 4 compared to the last two decades, the economic performance of the ECOWAS sub-region has improved significantly primarily due to increased trade activities (OSAbUOHIEN et al., 2019). With the normality of data, the jarque-bera test result rejected the null hypothesis of normal series distribution because all the variables were statistically significant at 1% with exception of Exportsij, which was statistically significant at 5%.

Cross-sectional dependency test and Panel unit root test
Cross-section dependence has to do with the impact of shocks in one country on another country when both countries belong in the panel data set (DE HOYOS & SARAFIDIS, 2006). The cross-sectional dependence was analyzed using PESARAN CD, PESARAN Scaled LM, and breusch-pagan LM tests (PESARAN, 2020) as is shown in table 6. However, the PESARAN CD test failed to reject the null hypothesis, indicating that there is no cross-sectional dependence.

Panel unit root test
To avoid spurious regression which may lead to wrong forecast, Three-panel unit root tests; Augmented Dicky Fuller (ADF-Fisher Chi-square), Levin, Lin Chu (LLC), and Philip perron (PP-Fisher Chi-square) were conducted to check stationarity (PESARAN, 2012(PESARAN, , 2020. The test results are presented in table 7. The table showed that all the variables are statistically significant and stationary at first difference which implies that all the variables are integrated in order (I(I)).
from the ECOWAS region. Whilst these results are synonymous with other findings, the magnitude of influence differs in this current study. For instance, an increase in the number of Confucius institutes will cause a 0.39% increase in trade volume whereas access to the sea route and WTO accounted for 0.47% and 0.77% increase in ECOWAS exports respectively. Similar to SUN, HUANG, & YANG (2014) analyses of China imports, the GDP of China (GDPi) will impede the volume of ECOWAS exports (-2.59%) to China since larger economies are more attracted to trade with their counterparts than weaker economies which explain China's high imports from America, Canada, Russia, and brazil than the African region. On the contrary, VON ESSEN (2017) revealed that agricultural trade flow from Sub-Saharan Africa region to China is enhanced by the GPD's of both economies.
Moreover, the volume of exports is negatively influenced by the level of infrastructural development in ECOWAS (InfraSj), geographical distance between ECOWAS-China (VON ESSEN, 2017;YANG et al., 2020;ZHANG & LI, 2009) (Dij), and the exchange rate of both partners (EXCij) (GUAN & IP PING SHEONG, 2020). A unit increase in infrastructural development will significantly decrease the volume of trade by -1.474 percent. Currently, there are few Agricultural-manufacturing industries therefore exports from the region are mainly unprocessed raw agricultural materials with a perishability rate, which are difficult to transport via long-distance sea route to China. With the gradual industrialization growth in Africa, most raw materials will be processed and exported to other closer regions like Europe, which is fairly closer to most ECOWAS countries than China. Additionally, the population of China, which has shaped china's food trade and consumption pattern for the past decades, was also found significant in this study (LIU & WANG, 2018;ZENG et al., 2021). The results suggested that China's population would significantly account for a 4.03% rise in the volume of Agricultural exports from the ECOWAS sub-region. Finally, our results backed the evidence that the ECOWAS region has mainly relied on intensive manual labor force for most agricultural production processes; therefore, its population serves as a driving factor for the growth and development of agricultural productivity and exports, which will lead to a 0.58% rise in the volume of exports.

CONCLUSION
In an attempt to unravel the what's and why's of ECOWAS-China Agricultural products trade, the above empirical analysis led to the following conclusion; (1) Although, the findings demonstrate the existence of bilateral trade between ECOWAS and China, the high cost of transporting goods because of the geographical distance between ECOWAS and China serves as an impediment to trade flow. Whist this results aligns with trade literature, compared to most European ports' proximity to most African states, it takes approximately 15 days more from China's closest seaport to the nearest ECOWAS country, which limits the possibility of a trade. The PPML results revealed that a unit increase in distance may lead to a decline in trade volume by -4.45%. additionally, China's GDP is negatively significant which suggested that trade volume will decline by -2.59 for a unit increase in Economic growth (GDP) since larger economies trade with each other, it is not surprising that China focuses more on trading with the USA, Australia, Russia, and other larger states.
Similar to other studies, the level of Infrastructural development greatly influences trade  Abdullahi et al. volume. Characterized by weak processing and manufacturing industries, African States account for the highest primary Agricultural exports to wealthier regions like the USA, Europe, and China who are well endowed to further process into furnished goods. From the regression results, A unit change (improvement) in the infrastructural growth of ECOWAS states will likely decrease the exports of the Agricultural products by (-1.47%). This phenomenon provides a key to why China trades with ECOWAS; although, barriers such as the volume of products and distance are currently not favorable.
(3) With one of the fastest middle-income earners population growth, China's population explains why ECOWAS Agricultural exports make way to the Chinese market despite the stifling trade barriers. The finding suggested that a unit increase in China's population will consequently translate to increased trade volume by 4.03%. This result meets the simple demandsupply assumption in that, a growing population will require equivalent growing food supplies to feed individuals and industrials; however, with limited arable lands in China, the ECOWAS regions remain a potential spot for supplementary agricultural raw materials.
However, an increase in the population of ECOWAS will hamper the volume of exports to China.
The impact of trade openness facilitates negotiations and positive mutual agreements between trade partners. In previous studies, China's accession to WTO has shown a positive effect on both volume, the number of trade partners, and products traded. In this current study, similar conclusions were reached with a positive (0.77%) and significant (0.0003) effect of WTO membership of both partners on the volume of trade.
(5) More uniquely, the number of Chinese Confucius institutes in ECOWAS countries play a positive and significant role as far as Agricultural trade between the two partners is concerned. This forms a foundation for further trade growth since the common language remains one of the key fulcrums of bilateral trade as purported in several studies. We concluded that the growing number of Confucius institutions in most ECOWAS countries accounts for improved negotiations hence growth in trade volume.

Policy implication
The policy proposals presented below aimed to guarantee that ECOWAS Agricultural trade ties with  China are not influenced by resource-seeking goals, but rather by a mutually beneficial relationship that aligns with the objectives of the ECOWAS regional trade block: Stimulating Agricultural trade growth through Strategic direction of FDI Contrary to the forms of Agricultural products exported to the European market and the United States, The level of primary products from ECOWAS to China are mainly limited to unprocessed commodities which are difficult and costly to transport. Since distance increases trade cost and consequently affects the volume of trade, Chinese direct investments towards ECOWAS should be directed towards upgrading Agro-industries to increase the manufacturing of semi and processed agro commodities that meet China's growing dynamic demand. This will enhance and expand the scope and volume of Agricultural trade whilst contribute towards job creation, the rapid transformation of the Agricultural industries; and consequently economic growth.

Intensifying trade associations and cooperation forums for win-win Economic benefits.
Again since trade associations such as WTO have been proven to positively enhance Agricultural trade, we suggested similar impacts to be derived from ongoing China-Africa trade negotiations and cooperation agreements. FOCAC and China-Africa Agricultural cooperation represent such forums where fair deals on Agricultural trade development may be enhanced in exchange for industrial and economic development.

Capitalizing on resource advantage and reversing challenges for trade growth
Finally, ECOWAS State should capitalize on the arable land size, which has not received the needed attention and investments though a positive driver of Agricultural exports. We there propose that to derive optimum economic benefits from favorable arable land sizes in ECOWAS States, prevailing challenges such as poor irrigation, road network, technical expertise, low level of research and technology and government support systems should be given maximum attention to intensify production levels and export volumes.

DECLARATION OF CONFLICT OF INTERESTS
The authors declare no conflict of interest.

AUTHORS' CONTRIBUTIONS
Conceptualization: bNA, MD and FKY. Data acquisition: bNA, FKY and HYI. Design of methodology and data analysis: bNA, FKY and MD. FKY and bNA prepared the draft of the manuscript. All authors critically revised the manuscript and approved of the final version.